Horton

Re: Skipping logical replication transactions on subscriber side

Masahiko Sawada <sawada.mshk@gmail.com>

View thread View on postgr.es
From: Masahiko Sawada <sawada.mshk@gmail.com>
To: "houzj.fnst@fujitsu.com" <houzj.fnst@fujitsu.com>
Cc: Alexey Lesovsky <lesovsky@gmail.com>, Peter Eisentraut <peter.eisentraut@enterprisedb.com>, PostgreSQL-development <pgsql-hackers@postgresql.org>, Amit Kapila <amit.kapila16@gmail.com>
Date: 2021-08-02T02:36:12Z
Lists: pgsql-hackers 
On Fri, Jul 30, 2021 at 3:47 PM houzj.fnst@fujitsu.com
<houzj.fnst@fujitsu.com> wrote:
>
> On July 29, 2021 1:48 PM Masahiko Sawada <sawada.mshk@gmail.com> wrote:
> >
> > Sorry I've attached wrong ones. Reattached the correct version patches.
>
> Hi,
>
> I had some comments on the new version patches.

Thank you for the comments!

>
> 1)
>
> -       relstate = (SubscriptionRelState *) palloc(sizeof(SubscriptionRelState));
> -       relstate->relid = subrel->srrelid;
> +       relstate = (SubscriptionRelState *) hash_search(htab, (void *) &subrel->srrelid,
> +                                                       HASH_ENTER, NULL);
>
> I found the new version patch changes the List type 'relstate' to hash table type
> 'relstate'. Will this bring significant performance improvements ?

For pgstat_vacuum_stat() purposes, I think it's better to use a hash
table to avoid O(N) lookup. But it might not be good to change the
type of the return value of GetSubscriptionNotReadyRelations() since
this returned value is used by other functions to iterate over
elements. The list iteration is faster than the hash table’s one. It
would be better to change it so that pgstat_vacuum_stat() constructs a
hash table for its own purpose.

>
> 2)
> + * PgStat_StatSubRelErrEntry represents a error happened during logical
>
> a error => an error

Will fix.

>
> 3)
> +CREATE VIEW pg_stat_subscription_errors AS
> +    SELECT
> +   d.datname,
> +   sr.subid,
> +   s.subname,
>
> It seems the 'subid' column is not mentioned in the document of the
> pg_stat_subscription_errors view.

Will fix.

>
>
> 4)
> +
> +                   if (fread(&nrels, 1, sizeof(long), fpin) != sizeof(long))
> +                   {
>     ...
> +                   for (int i = 0; i < nrels; i++)
>
> the type of i(int) seems different of the type or 'nrels'(long), it might be
> better to use the same type.

Will fix.

Regards,

-- 
Masahiko Sawada
EDB:  https://www.enterprisedb.com/

Commits

  1. Test ALIGNOF_DOUBLE==4 compatibility under ALIGNOF_DOUBLE==8.

  2. Reorder subskiplsn in pg_subscription to avoid alignment issues.

  3. Add ALTER SUBSCRIPTION ... SKIP.

  4. Optionally disable subscriptions on error.

  5. Update docs of logical replication for commit 8d74fc96db.

  6. Respect permissions within logical replication.

  7. Fix regression test failure caused by commit 8d74fc96db.

  8. Add a view to show the stats of subscription workers.

  9. Add logical change details to logical replication worker errcontext.

  10. Rename LOGICAL_REP_MSG_STREAM_END to LOGICAL_REP_MSG_STREAM_STOP.

  11. Fix typo in protocol.sgml.

  12. Remove unused argument in apply_handle_commit_internal().

  13. Fix replication of in-progress transactions in tablesync worker.

  14. Reorder pg_sequence columns to avoid alignment issue